Abstract

Wing rock phenomenon is manifested by a limit cycle oscillation predominantly in roll about the body axis. This self-induced rolling oscillation is highly annoying to the pilot and poses serious limitation to the combat effectiveness. The maneuvering envelope of an aircraft exhibiting this behavior is also seriously restricted because the maximum angle of attack (AOA) is often limited by the onset of wing rock before the occurrence of stall. This thesis deals with the control and simulation investigation of wing rock phenomenon by the five new control schemes for the different assumptions of wing rock model. First, a variable phase control scheme, based on the results of energy analysis of wing rock hysteresis, is developed for wing rock suppression. Its main advantage is small control power need. For the tracking control of wing rock at a fixed AOA, fuzzy PD control is then designed; however, if wing rock suffers from external disturbance, fuzzy PD control exhibits a big tracking error. To overcome this disadvantage, variable universe fuzzy PD control is proposed to achieve the precise tracking control. For the tracking control of wing rock with a time-varying AOA and uncertainties, the NDOFEL scheme, a nonlinear disturbance observer (NDO) combined with a feedback-error-learning (FEL) strategy, is proposed for a class of time-varying nonlinear systems with unknown disturbances, where the nominal model of wing rock control is assumed available. The proposed NDOFEL not only extends the NDO into time-varying nonlinear systems but also improves the precision of tracking control. Finally, because aircraft at a high AOA operate in nonlinear flight regimes in which the dynamics are very complex, aircraft's wing rock model is usually unavailable. A reinforcement adaptive fuzzy control scheme is presented to guarantee the stability of the closed-loop system and the convergence of the tracking error. The proposed control schemes emphasize stability, robustness, and simplicity such that they can be applied for on-line learning and real-time control. Numerical cases in each scheme are used to confirm the effectiveness and robustness of the proposed schemes.